Going away after a few initial dabs doesn't sound like heat fade.
Have you tried to touch the drum after the first failed stop? The temperature at the lining surface (drum OD) would be above 200° F if it's heat.
I suspect you don't have good contact, all the comments on skimming etc. are spot on.
Venting the drum is useful for a few reasons:
1. makes you feel like you accomplished something (rider confidence = good!)
2. frightens the other riders (competing rider confidence = bad!)
3. reduces unsprung weight
4. reduces temp build-up
However, this isn't as effective as it appears, and yes the factory(ies) haven't always done their homework on this.
Air only flows from a high pressure area to a low pressure area.
Given the picture: which is which?
The answer: there may not be any, the pressure differential may only occur with changes in air flowing over the bike, and even reverse at speeds.
The ideal is to pull air in centrally and exit through holes in the drum OD. The entry can be anywhere in the backing plate, but if the plate is largely masked by the swing-arm, etc. you'll need a horn extending from the plate into the airstream (like the 1971-72 BSA front brake). To prevent the plate from cracking, several holes in an arc or a pattern of 1/8" holes in the same spot can be fed by a single horn, which can be secured by weld, small screws, etc.
For the drum not to simply crack or distort, very little of the rim structure can be removed, and that very carefully. The holes have to be quite small, and widely spaced from each other, from the outer edge, and the radius where the rim joins the web.
How many holes? If the rim is 1" wide, the usable width may be only 3/4" (1/8" safety on either side). A 7" OD drum by 1" wide has 16.5 square inches of available friction surface.
A single 1.5" entry (1.77 square inch area) will make quite a difference in internal air temperature, especially if it's pressurized.
Dividing this into 5/32" (.15625" or 4mm) rim holes: alternating 1 across and 2 across (to make shoe contact discontinuous) times 40 rows (easy to do - line them up with the spokes), over 1/2" apart on the circumference, makes 60 holes, total area 1.15 square inches, about the same as a 1-3/16" hole (and only 5.2% of the rim surface, leaving 94.8% for structure).
I've seen comments that the holes at the edge should be closer (more numerous) than the holes at the radius by perhaps 1.5:1 or 2:1, since the edge is always hotter (it has no heat transfer by conduction to the drum body), but it's also the weakest area, so I'm not sure it's safe. The holes should be lightly counter-sunk then radius from both sides both to prevent cracks and improve air-flow. Since the drum interior is lightly pressurized, air will always from toward the largest exit (the drum OD), and carry heat away with it.
If the drum is vented on the vertical (web) surface, it's still a good idea to use a horn entry for the backing plate since otherwise air-flow past the 2 sides may be very similar, result: no air through the drum.
Going to the pictured plate with holes and screens: a simple yarn test will show you if you have air circulation - at high speed the tuft positioned at one set of holes will pull straight back, and the other will flutter. If both behave alike flow is minimal.
If you want minimal change in thickness (no extended horn), a simple angled ramp out of folded 1/32" aluminum can be secured over the screen. Point one forward (high pressure), and one backward (induced vacuum at the open for low pressure), although I suspect that the forward-facing scoop does most of the work.
